SQL Server pivot values with different data types - sql

Question

i am trying to pivot all values in different type in MSSQL 2016. I could not find a way how i can pivot different data types..
The first table are initial form / structure. The second table is the desired shape.
I was trying the following SQL code to pivot my values
SELECT
[id] AS [id],
FIRSTNAME,
LASTNAME,
BIRTHDATE,
ADDRESS,
FLAG,
NUMBER
FROM (
SELECT
[cm].[key] AS [id],
[cm].[column] AS [column],
[cm].[string] AS [string],
[cm].[bit] AS [bit],
[cm].[xml] AS [xml],
[cm].[number] AS [number],
[cm].[date] AS [date]
FROM [cmaster] AS [cm]
) AS [t]
PIVOT (
MAX([string]) --!?!?
FOR [column] IN (
FIRSTNAME,
LASTNAME,
BIRTHDATE,
ADDRESS,
FLAG,
NUMBER
)
) AS [p]

Answer 1

I think your best bet is to use conditional aggregation, e.g.
SELECT cm.id,
FIRSTNAME = MAX(CASE WHEN cm.[property] = 'firstname' THEN cm.[string] END),
LASTNAME = MAX(CASE WHEN cm.[property] = 'lastname' THEN cm.[string] END),
BIRTHDATE = MAX(CASE WHEN cm.[property] = 'birthddate' THEN cm.[date] END),
FLAG = CONVERT(BIT, MAX(CASE WHEN cm.[bit] = 'flag' THEN CONVERT(TINYINT, cm.[boolean]) END)),
NUMBER = MAX(CASE WHEN cm.[property] = 'number' THEN cm.[integer] END)
FROM cmaster AS cm
GROUP BY cm.id;
Although, as you can see, your query becomes very tightly coupled to your EAV model, and why EAV is considered an SQL antipattern. Your alternative is to create a single column in your subquery and pivot on that, but you have to convert to a single data type, and lose a bit of type safety:
SELECT id, FIRSTNAME, LASTNAME, BIRTHDATE, ADDRESS, FLAG, NUMBER
FROM (
SELECT id = cm.[key],
[column] = cm.[column],
Value = CASE cm.type
WHEN 'NVARCHAR' THEN cm.string
WHEN 'DATETIME' THEN CONVERT(NVARCHAR(MAX), cm.date, 112)
WHEN 'XML' THEN CONVERT(NVARCHAR(MAX), cm.xml)
WHEN 'BIT' THEN CONVERT(NVARCHAR(MAX), cm.boolean)
WHEN 'INT' THEN CONVERT(NVARCHAR(MAX), cm.integer)
END
FROM cmaster AS cm
) AS t
PIVOT
(
MAX(Value)
FOR [column] IN (FIRSTNAME, LASTNAME, BIRTHDATE, ADDRESS, FLAG, NUMBER)
) AS p;

Answer 2

In order to make the result as per your request, first thing is we need to bring the data in to one format which is compatible with all data types. VARCHAR is ideal for that. Then prepare the base table using a simple select query, then PIVOT the result.
In the last projection, if you want, you can convert the data back in to the original format.
This query can be written dynamically as well to obtain the result as records are added. Here I provide the static answer according to your data. If you need a more generic dynamic answer, let me know. So I can post here.
--data insert scripts I used:
CREATE TABLE First_Table
(
[id] int,
[column] VARCHAR(10),
[string] VARCHAR(20),
[bit] BIT,
[xml] [xml],
[number] INT,
[date] DATE
)
SELECT GETDATE()
INSERT INTO First_Table VALUES(1, 'FIRST NAME', 'JOHN' , NULL, NULL, NULL, NULL)
INSERT INTO First_Table VALUES(1, 'LAST NAME', 'DOE' , NULL, NULL, NULL, NULL)
INSERT INTO First_Table VALUES(1, 'BIRTH DATE', NULL , NULL, NULL, NULL, '1985-02-25')
INSERT INTO First_Table VALUES(1, 'ADDRESS', NULL , NULL, 'SDFJDGJOKGDGKPDGKPDKGPDKGGKGKG', NULL, NULL)
INSERT INTO First_Table VALUES(1, 'FLAG', NULL , 1, NULL, NULL, NULL)
INSERT INTO First_Table VALUES(1, 'NUMBER', NULL , NULL, NULL, 20, NULL)
SELECT
PIVOTED.* FROM
(
--MAKING THE BASE TABLE FOR PIVOT
SELECT
[id]
,[column] AS [COLUMN]
, CASE WHEN [column] = 'FIRST NAME' then [string]
WHEN [column] = 'LAST NAME' then [string]
WHEN [column] = 'BIRTH DATE' then CAST([date] AS VARCHAR(100))
WHEN [column] = 'ADDRESS' then CAst([xml] as VARCHAR(100))
WHEN [column] = 'FLAG' then CAST([bit] AS VARCHAR(100))
else CAST([number] AS VARCHAR(100)) END AS [VALUE]
FROM First_Table
) AS [P]
PIVOT
(
MIN ([P].[VALUE])
FOR [column] in ([FIRST NAME],[LAST NAME],[BIRTH DATE],[ADDRESS],[FLAG],[NUMBER])
) AS PIVOTED
RESULT:

Answer 3

SQL：
SELECT
            ID,
            FIRSTNAME,
            ...,
            FLAG = CAST (FLAG AS INT),
            ...
FROM
            (
            SELECT
                        *
            FROM
                        (
                        SELECT
                                    f.ID,
                                    f.PROPERTY,
                                    f.STRING + f."INTEGER" + f.DATETIME + f.BOLLEAN + f.XML AS COLS
                        FROM
                                    FIRSTTBL f)
            PIVOT(
                        min(COLS) FOR PROPERTY IN
                                    (
                                    'firstname' AS firstname,
                                    'lastname' AS lastname,
                                    'birthdate' AS birthdate,
                                    'address' AS address,
                                    'flag' AS flag,
                                    'number' AS "NUMBER"
                                    )
                        )
            )
According to the original table, there is one and only one non-null value among STRING, INTEGER, DATETIME, BOLLEAN and XML columns for any row, so we just need to get the first non-null value and assign it to the corresponding new column. It is not difficult to perform the transposition using PIVOT function, except that we need to handle different data types according to the SQL rule, which requires that each column have a consistent type. For this task, first we need to convert the combined column values into a string, perform row-to-column transposition, and then convert string back to the proper types. When there are a lot of columns, the SQL statement can be tricky, and dynamic requirements are even hard to achieve.
Yet it is easy to write the code using the open-source esProc SPL:
　
A
1
=connect("MSSQL")
2
=A1.query#x("SELECT * FROM FIRSTTBL")
3
=A2.pivot(ID;PROPERTY,~.array().m(4:).ifn();"firstname":"FIRSTNAME", "lastname":"LASTANME","birthdate":"BIRTHDAY","address":"ADDRESS","flag":"FLAG","number":"NUMBER")
SPL does not require that data in the same column have consistent type. It is easy for it to maintain the original data types while performing the transposition.